Title :
A novel method for boundary detection in low contrast images
Author :
Mishra, Saurabh Tarun ; Ali, Imran
Author_Institution :
KIIT Univ., Bhubaneswar, India
Abstract :
Segmentation of region of interest from image is the most crucial step in field of machine vision for extracting features of object from images. In contrast of this field we are proposing one model which have greater time computational efficiency as compared to existing models. Our algorithm is derived from idea to divide and conquer, we are dividing our image into different small coordinates. And if the pixels of image satisfy the threshold it would be the points on region. Our model work on both low contrast images and videos with illumination variation with greater time efficiency. Our proposed algorithm proper detection of contour when an object stays still in the frame and become a background.
Keywords :
computer vision; feature extraction; image segmentation; object detection; boundary detection; contour detection; feature extraction; illumination variation; image segmentation; low contrast images; machine vision; region of interest segmentation; Algorithm design and analysis; Approximation algorithms; Clocks; Clustering algorithms; Image edge detection; Image segmentation; MATLAB; Background subtraction; boundary detection; centroid; partition of image;
Conference_Titel :
Communications and Signal Processing (ICCSP), 2014 International Conference on
Conference_Location :
Melmaruvathur
Print_ISBN :
978-1-4799-3357-0
DOI :
10.1109/ICCSP.2014.6949994
